車輛定位資訊為車輛導航、先進駕駛輔助系統(Advanced Driver Assistance System,簡稱ADAS)與自動駕駛技術的關鍵資訊，且準確的車輛定位是實施新興智能交通系統最重要的前提之一，全球定位系統(global positioning system，簡稱GPS)可以快速量測到車輛的絕對座標，其常被應用於車輛的定位，由於GPS會受到一些因素的影響(例如：星曆誤差、衛星鐘差和多路徑效應)，車輛接收到的GPS座標通常伴有很大的誤差，這會危害自動駕駛車輛的安全性，為了解決這個問題，已經有許多廣泛的合作式車輛定位系統的研究，其可以與相鄰車輛相互交換訊息，甚至可以將GPS結合多種傳感器(sensor)以提高定位效能，現存基於規則的定位方法其權重設計能力較差，這將導致多個傳感器之間的關係無法得到充分表達，因此，基於深度學習的非線性和多層隱藏層的特性，使其在權重設計方面有較好的表現，於是本論文引入了深度學習方法以獲得更好的定位結果。
本論文不僅通過多層感知器(multi-layer perceptron，簡稱MLP)充分提取了傳感器特徵，還進一步整合了多時隙的資訊以提升定位效能，然而，現存基於規則的定位方法無法針對駕駛行為進行數學式的建模，因此，尚未有整合多時隙資訊以進一步提升定位效能的演算法被提出，而在整合多時隙資訊方面，本論文引入了長短期記憶(long short-term memory，簡稱LSTM)以更好地處理具時序性關係的資料，此外，為了更進一步提高車輛定位效能，本論文引入了圖卷積網路(graph convolution network，簡稱GCN)，以同時考慮時間及空間上的相關性特徵，模擬結果將會證實所提出架構的優越效能。

Vehicle positioning information is the key information for vehicle navigation、 advanced driver assistance system (ADAS) and autonomous driving technology. The global positioning system (GPS) is commonly used in vehicle localization because GPS can collect the absolute coordinate of vehicles rapidly. Owing to GPS suffer from some noise affect(ex：ephemeris error、satellite clock bias and multipath effect), we usually receive the GPS coordinate of the vehicle with large error, it harm the safety of self-driving vehicles. Extensive research has been developed to tackle with this problem. There exist many cooperative vehicle localization methods, which exchange information with neighboring vehicles and even integrated GPS with various sensors to improve localization performance. Existing rule-based methods has poor weighting design capability; it will lead to the relationship between multiple sensors cannot be fully expressed. Thus, we introduced deep learning (DL) based approach to get better localization results, due to the nonlinear and complex weighting design process of deep learning.
We not only fully extracted sensors information by multi-layer perceptron (MLP), but further integrated multi-time slot information to improve the localization performance. However, the existing rule-based localization approach cannot perform driving behavior modeling. Therefore, no algorithm has been proposed to further improve the localization performance based on multi-time slot information. In terms of integrating multi-time slot information, we introduced long short-term memory (LSTM) which is a better model to deal with data with time-series relationships. In addition, to further improve vehicle localization performance, the thesis introduced graph convolution network(GCN) to consider both time and space correlation simultaneously. Simulation results will confirm the superior performance of the proposed architecture.

[1] J. V. Brummelen, M. O’Brien, D. Gruyer, and H. Najjaran, “Autonomous vehicle perception: The technology of today and tomorrow,” Transp. Res. C Emerg. Technol., vol. 89, no. 1, pp. 384–406, Apr. 2018. [Online] Available: https://www.sciencedirect.com/science/ article/pii/S0968090X18302134
[2] E. C. Eze, S. Zhang, and E. Liu, “Vehicular ad hoc networks (VANETs): Current state, challenges, potentials and way forward,” in Proc. Int. Conf. Autom. Comput., pp. 176–181, Sep. 2014.
[3] S. Kuutti, S. Fallah, K. Katsaros, M. Dianati, F. Mccullough, and A. Mouzakitis, “A survey of the state-of-the-art localization techniques and their potentials for autonomous vehicle applications,” IEEE Internet Things J., vol. 5, no. 2, pp. 829–846, Apr. 2018.
[4] K. Liu, H. B. Lim, E. Frazzoli, H. Ji, and V. Lee, “Improving positioning accuracy using GPS pseudorange measurements for cooperative vehicular localization,” IEEE Trans. Veh. Technol., vol. 63, no. 6, pp. 2544–2556, Jul. 2014.
[5] M. A. Hossain, I. Elshafiey and A. Al-Sanie, “High accuracy GPS-free vehicular positioning based on V2V communications and RSU-assisted DOA estimation, ” 2017 9th IEEE-GCC Conf. and Exhibition, Manama, pp. 1-5.
[6] M. Rohani, D. Gingras, and D. Gruyer, “A novel approach for improved vehicular positioning using cooperative map matching and dynamic base station DGPS concept,” IEEE Trans. Intell. Transp. Syst., vol. 17, no. 1, pp. 230–239, Jan. 2016.
[7] S. Nam, D. Lee, J. Lee and S. Park, “CNVPS: Cooperative neighboring vehicle positioning system based on vehicle-to-vehicle communication,” in IEEE Access, vol. 7, pp. 16847-16857, 2019.
[8] R. Parker and S. Valaee, “Vehicle localization in vehicular networks,” in Proc. 64th IEEE VTC-Fall, pp. 1–5, 2006.
[9] N. Drawil and O. Basir, “Toward increasing the localization accuracy of vehicles in VANET,” in Proc. IEEE ICVES, pp. 13–18, 2009.
[10] J. Yao, A. Balaei, M. Hassan, N. Alam, and A. Dempster, “Improving cooperative positioning for vehicular networks,” IEEE Trans. Veh. Technol., vol. 60, no. 6, pp. 2810–2823, Jul. 2011.

[11] F. d. P. Müller, E. M. Diaz and I. Rashdan, “Cooperative positioning and radar sensor fusion for relative localization of vehicles,” IEEE Intell. Veh. Symp., Gothenburg, pp. 1060-1065, 2016.
[12] G. Mao, B. Fidan, and B. D. O. Anderson, “Wireless sensor network localization techniques,” Comput. Netw., vol. 51, no. 10, pp. 2529–2553, Jan. 2007.
[13] S. B. Cruz, T. E. Abrudan, Z. Xiao, N. Trigoni, and J. Barros, “Neighbor-aided localization in vehicular networks,” IEEE Trans. Intell. Transp. Syst., 2017.
[14] L. Altoaimy and I. Mahgoub, ‘‘OWL: Optimized weighted localization for vehicular ad hoc networks,’’ in Proc. Int. Conf. Connected Vehicles Expo, pp. 699–704, Nov. 2014.
[15] J. Li, N. Song, M. Li, Q. Cai, and G. Yang, ‘‘Improving positioning accuracy of vehicular navigation system during GPS outages utilizing ensemble learning algorithm,’’ Inf. Fusion, vol. 35, pp. 1–10, May 2017.
[16] Z. Zhao et al., ‘‘LSTM network: A deep learning approach for short-term traffic forecast,’’ IET Intell. Transp. Syst., vol. 11, no. 2, pp. 68–75, 2017.
[17] K. He, X. Zhang, S. Ren, and J. Sun, ‘‘Delving deep into rectifiers: Surpassing human-level performance on imagenet classification,’’ in Proc. IEEE Int. Conf. Comput. Vis., pp. 1026–1034, Dec. 2015.
[18] Z. Zhang, P. Cui and W. Zhu, ‘‘Deep learning on graphs: A survey,’’ IEEE Trans. Knowl. Data Eng., Mar. 2020.
[19] T. N. Kipf and M. Welling. ‘‘Semi-supervised classification with graph convolutional networks.’’, 2016, [Online]. Available: https://arXiv:1609.02907
[20] D. K. Hammond, P. Vandergheynst, and R. Gribonval, ‘‘Wavelets on graphs via spectral graph theory,’’ Appl. Comput. Harmon. Anal., vol. 30, no. 2, pp. 129–150, Mar. 2011.
[21] K. Guo et al., ‘‘Optimized graph convolution recurrent neural network for traffic prediction,’’ IEEE Trans. Intell. Transp. Syst., Jan. 2020.
[22] J. Bruna, W. Zaremba, A. Szalm, and Y. LeCun, ‘‘Spectral networks and locally connected networks on graphs,’’ in Proc. Int. Conf. Learn. Represent., pp. 1-14, 2013.
[23] M. Defferrard, X. Bresson and P. Vandergheynst, ‘‘Convolutional neural networks on graphs with fast localized spectral filtering’’, in Proc. 28th Int. Conf. Neural Inf. Process. Syst., pp. 3837-3845, 2016.

[24] D. K. Hammond, P. Vandergheynst, and R. Gribonval, ‘‘Wavelets on graphs via spectral graph theory,’’ Appl. Comput. Harmon. Anal., vol. 30, no. 2, pp. 129–150, Mar. 2011.
[25] Dedicated Short Range Communications (DSRC) Message Set Dictionary, SAE Standard J2735_200911, DSRC Committee, 2009.